chore(bench): exclude short rides

3 files changed, 19 insertions, 7 deletions

diff --git a/cli/cli.go b/cli/cli.go
index d3aff22..4459946 100644
--- a/cli/cli.go
+++ b/cli/cli.go
@@ -32,6 +32,8 @@ var (
 	backtest    = flag.Bool("backtest", false, "run data benchmarks comparing computed values against stored reference values")
 )
 
+const backtestMinimumElapsedTimeS = 15 * 60
+
 func Run(db *sql.DB, configPath string, appConfig config.Config) {
 	switch {
 	case *download:
@@ -84,7 +86,7 @@ func runBacktest(db *sql.DB) error {
 		if i > 0 && i%50 == 0 {
 			slog.Info("Backtest progress", "processed", i, "total", len(items))
 		}
-		if item.GPXPath == "" {
+		if !isBacktestEligible(item) {
 			continue
 		}
 		parsed, err := ride.ParseFile(ride.GPXRideParser{}, item.GPXPath)
@@ -93,9 +95,6 @@ func runBacktest(db *sql.DB) error {
 			continue
 		}
 		reference := time.Duration(item.MovingTimeS) * time.Second
-		if reference <= 0 {
-			continue
-		}
 		computed := ride.MovingTime(parsed)
 		samples = append(samples, backtestSample{
 			externalID:  item.ExternalID,
@@ -135,10 +134,14 @@ func runBacktest(db *sql.DB) error {
 		fmt.Fprintf(table, "p%g\t%.1f\n", p, distribution[p])
 	}
 	table.Flush()
-	slog.Info("Moving time backtest", "rides", len(samples), "mean_err_pct", math.Round(meanAbsErr(absErrors)*10)/10, "stderr_pct", math.Round(stdErrAbsErr(absErrors)*10)/10)
+	slog.Info("Moving time backtest", "rides", len(samples), "min_elapsed_s", backtestMinimumElapsedTimeS, "mean_err_pct", math.Round(meanAbsErr(absErrors)*10)/10, "stderr_pct", math.Round(stdErrAbsErr(absErrors)*10)/10)
 	return nil
 }
 
+func isBacktestEligible(item rides.Ride) bool {
+	return item.GPXPath != "" && item.ElapsedTimeS >= backtestMinimumElapsedTimeS && item.MovingTimeS > 0
+}
+
 func meanAbsErr(absErrors []float64) float64 {
 	var sum float64
 	for _, err := range absErrors {
diff --git a/cli/cli_test.go b/cli/cli_test.go
index 8290bce..ae59c6c 100644
--- a/cli/cli_test.go
+++ b/cli/cli_test.go
@@ -3,9 +3,18 @@ package cli
 import (
 	"testing"
 
+	"github.com/martinlehoux/biking_home/rides"
 	"github.com/stretchr/testify/assert"
 )
 
+func TestBacktestEligibilityExcludesShortRides(t *testing.T) {
+	short := rides.Ride{GPXPath: "ride.gpx", ElapsedTimeS: 15*60 - 1, MovingTimeS: 1}
+	minimum := rides.Ride{GPXPath: "ride.gpx", ElapsedTimeS: 15 * 60, MovingTimeS: 1}
+
+	assert.False(t, isBacktestEligible(short))
+	assert.True(t, isBacktestEligible(minimum))
+}
+
 func TestBacktestStatsUseUnroundedSamples(t *testing.T) {
 	absErrors := []float64{0, 0.1, 0.1}
 
diff --git a/ride/moving_time.go b/ride/moving_time.go
index c98e6b1..9ac6c05 100644
--- a/ride/moving_time.go
+++ b/ride/moving_time.go
@@ -16,8 +16,8 @@ const movingTimeMinPowerW = 20
 // that can count as moving. Longer gaps are pauses (stops, lost signal).
 const movingTimeMaxGap = 120 * time.Second
 
-// Bench (356 rides, `-backtest`):
-// absMean 2.4%, stderr 0.2, p50 0.7, p75 2.9, p90 6.7, p95 9.6, p99 19.9.
+// Bench (237 rides, `-backtest`, elapsed time >= 15 minutes):
+// absMean 2.4%, stderr 0.3, p50 0.6, p75 2.9, p90 7.1, p95 10.7, p99 19.9.
 func MovingTime(r Ride) time.Duration {
 	return movingTime(r, movingTimeMinSpeedMps, movingTimeMaxGap)
 }